Impact of the Ownership Structure on the Corporate Diversification via Acquisitions: Empirical Post-Crisis Evidence from European Industrial Companies

Impact of the Ownership Structure on the Corporate Diversification via Acquisitions: Empirical Post-Crisis Evidence from European Industrial Companies

Master Thesis in Finance, January 2018

Nikita Sommer

*, supervised by Dr. V. Purice

a

Faculty of Economics and Business, University of Groningen, The Netherlands

ARTICLE INFO

JEL classification:

G32 G34

Keywords:

Ownership structure Diversification Blockholders Agent theory Acquisitions

ABSTRACT

Using panel data on 255 European exchange-listed, industrial companies, I examine the impact of the ownership structure on corporate diversification via acquisitions in the post-crisis period 2010-2016. It is shown that the liquidity of a company, which allows for managerial control, gives the management the chance to initiate acquisitions and further facilitates inorganic diversifica- tion. An exemption is one-directional diversification, which refers to either only sectoral or only geographical diversifying in this study. By differentiating between sizes and quantities of the shareholders, the paper shows that companies with large share- holders perform less acquisitions over companies without large shareholders. Furthermore, this effect is enhanced by multiple large shareholders. It is shown that companies, with high dispersi- ty in ownership by the shareholders´ geographical origins, acquire less companies when at least one shareholder holds more than 10% of the shares. However, the positive impact of management control on acquisitions is offset by a high level of dispersity.

✩ The author is profoundly grateful to Dr. V. Purice for the great guidance and helpful inputs

* Corresponding author at: University of Groningen, Faculty of Economics and Business, Groningen, The Netherlands. Student number: 3248917

Email: nikita.sommer@gmail.com (N. Sommer)

1. Introduction

Globalization is moving progressively, wherefore companies are urged to expand. There are multiple ways to control expansion and growth. One solution could be to increase their mar- ket share to reduce the number of competitors. Another solution could be to diversify to open new markets, which could be an expansion in the same sector but in a different country, in a different sector but the same country or even both directions - diversification into a new sec- tor in a new country. Commonly, growth is partly achieved inorganically via acquisitions (Shimizu et al., 2004) because organic growth may be too costly or time-consuming. In cer- tain cases, it even is not possible to access a foreign market through organic growth and ex- pansion (Shimizu et al., 2004). Besides the general reference to globalization, worldwide trends such as industry consolidation and privatization support the vast increase in mergers and acquisitions (M&A) activity.

The executive and supervisory boards play essential roles in the decision-making procedure towards the acquisition of a company. Therefore, the management and the large shareholder are crucial in this process. These parties´ motivations and incentives vastly determine the like- lihood to acquire and diversify.

The influence of ownership structure on firm´s performance has been researched extensively in the theoretical and empirical literature. A common focus of these studies was the impact of large shareholders and family ownership. Andres (2008) found from a sample of German ex- change-listed companies that family firms are more profitable than widely-held firms and they outperform with other types of blockholders, which is in line with other studies. Additionally, it is critical that the founding family is still active in the executive or supervisory board.

The influence of the ownership structure on the diversification behavior of the company got only some attention in the late 90s and early 2000s, which was mainly focused on the US market. Caprio et al. (2011) revived the topic again, focusing on the Continental European market. This study aims to amend this underrepresented field by not narrowing the focus on family control as well as using recent post-crisis data. In contrast to Caprio et al. (2011), this study focuses only on companies that have been involved in acquisitions as bidders and not as targets. This is consistent with the US focused research in this field (Basu et al., 2009; Bau- guess and Stegemoller, 2008; Klasa, 2007). Whereas most research focusses on the influence of family ownership, this study investigates the influence of different kinds of nature of the shareholders, their size of the shareholdings as well as their geographical origins.

The paper is structured as followed: Section 2 will review the relevant literature, section 3

will discuss the developing of hypotheses, section 4 will describe the data and methodology,

section 5 elaborate on empirical findings and section 6 will be the conclusion.

2. Literature

In this section, main theories related to corporate finance theory will be reviewed. More spe- cifically describing the implications on Mergers & Acquisitions (M&A) and the diversifica- tion behavior. Lastly, the impact of certain natures of owners on the acquisition and diversifi- cation behavior will be described.

2.1. Corporate Finance Theory

The early theories to explain diversification in the 90s were mainly based upon the resource based view (Robins and Wiersema, 1995) and transaction cost economics (Silverman, 1999;

Hautz et al. 2013). More concrete, they focused on economics of scope (Hitt, Hoskisson and Ireland, 1994), transfer learning (Hitt, Hoskisson and Kim, 1997) and organizational require- ments which limit diversification (Geringer, Tallman and Olson, 2000).

In more recent studies, the agency theory provides the theoretical foundations for research in this field (Denis et al., 1997). Agency theory assumes that managers, as agents, pursue their own personal goals rather than the interests of their principals, the company´s owners (Fama and Jensen, 1983). Build on the classic agency literature, having large shareholders mitigates the monitoring of managers in situations of dispersed ownership, since they have the incen- tive as well as the power to effectively monitor managers (Schleifer and Vishny, 1986, 1997;

Demsetz, 1983, 1986). A contrary argument is that large shareholders can lead to high agency costs in the case they align themselves with managers and use their power to expropriate wealth from minority shareholders (La Porta et al.). Large shareholders are typically classified as having 10% or more of the shareholdings (Attig et al., 2008; Laeven and Levine, 2008).

Latterly, the downside of higher agency costs has been stressed and researched assuming that low agency costs mean higher profitability, which was measured based on return on equity (RoE) and earnings before interests and taxes (EBIT)/total assets (Liu et al., 2014; Jiang et al., 2017).

Acquisitions need great amounts of cash in most cases and, it requires approval from many

organizational instances within the corporation, including the supervisory board. The power

distribution between the management and the supervisory board is consequently critical dur-

ing the decision-making process of whether to do acquisitions. According to Jean Tirole

(2006), the extent of managerial control increases with the strength of the balance sheet,

therefore leads to the assumption that a stronger balance sheet separates ownership and con-

trol further. The most feasible measurement of balance sheet strength in the theoretical model

by Tirole are the company´s amount of cash on hand and the market interest rate. A further

measure which is difficult to obtain, is the private benefit the entrepreneur would get from not

putting in effort into the work as well as the likelihood ratio of the entrepreneur to put in ef-

fort or not.

The relationship between the entrepreneur and the investor can usually be described as either arm´s-length relationship in the case of a strong balance sheet or otherwise relationship lend- ing according to Tirole (2006, 401). More in depth, the entrepreneur´s incentive scheme is positively influenced (“distressed”) by the strong balance sheet and thus the relationship is less risky for the investor, which further makes him ask for less payback. Therefore, the en- trepreneurial return increases and the managerial control decreases as stated in Tirole (2006, 388). The transfer of control rights to the investors, which is needed in cases of weak balance sheets to still perform acquisitions or generally spoken investments, increases the pledgeable income and further facilitates financing. This relationship theory bases on the Aghion-Bolton Model (Hart, 1995). Another reason for the separation of ownership and control can be that large minority shareholders often decide for the small group of representatives, which leads to a separation of ownership (formal control) and control (real control by the managers) (Tirole, 2006, 399), reflecting the partly contrary argumentation by Denis et al. (1997) and Tirole (2006),

Considering both views, according to Jiang et al. (2017), a recent literature has emerged pro- posing that an optimal compromise is to have multiple large shareholders to address both agency issues (Attig et al., 2008; Laeven and Levine, 2008; Bennedsen and Wolfenzon, 2000).

In that situation, they monitor the managers as well as each other because they have a high share of their wealth invested into the organization (further depending on the nature of owner, chapter 2.4). However, analyzing this constellation on the basis of the agency theory leads to the problem that the large shareholders may collude together against the small shareholders (LaPorta et al., 1999; Bennedsen and Wolfenzon, 2000). Furthermore, high collaboration and bargaining costs between large shareholders may arise (Gomes and Novaes, 2006). Based upon this common literature, Kumar and Zattoni (2015), the corporate governance literature usually hypothesizes a U-shaped relationship between ownership concentration and firm per- formance. Carrying forward this argumentation, the relationship between ownership concen- tration and acquisition behavior should be similar when assuming a dependency between firm performance and shareholder value.

So far, the empirical literature has found diverse results on the relationship between owner- ship concentration and acquisition behavior, which also depends on the region that research was based on (Konjin et al., 2011; Laeven and Levine, 2008). One line of argument is that if the large shareholder has a low level of control, the agency costs are high through the “en- trenchment effect

” (Jensen and Meckling, 1976; Claessens et al., 2002). Vice versa, if the large shareholder has a high level of ownership, the agency costs are low due to the “align-

1 The “entrenchment effect” describes the situation that the management wants to reduce their personal risk by diversification. In the case that the managerial control is high, the management therefore supports diversifying acquisitions.

ment-of-interest effect

” as the shareholder benefits more from reducing the agency costs.

Furthermore, large shareholders on the lower end (non-controlling shareholders) tend to align with management and use the control to expropriate wealth from minority shareholders, re- ferred to as the “free-rider problem” (Grossman and Hart, 1980; Holstrom, 1982). Controlling shareholders benefit the most from reducing agency costs because in that situation less acqui- sitions are enabled which would decrease the shareholder value (Jiang et al., 2017). For that reason, large shareholders are usually further divided into controlling and non-controlling (Jiang et al., 2017).

2.2. Rationales for Mergers and Acquisitions

Several factors have been identified to explain the occurrence of M&A and the effect of how they take place; examples include synergies (Andrade et al., 2001), managerial empire build- ing (Jensen, 1986), managerial hubris or overconfidence (Malmendier and Tate, 2008), and bidder overvaluation (Shleifer and Vishny, 2003). Parts of these arguments are overlapping with the general reasoning for (organic) diversification. However, another aspect is added due to the process of buying and often bidding; namely aspects from Behavioral Finance play more important roles. Since the shareholder structure of a firm shapes the decision maker's incentives, this further influences the decision-making process for acquisitions. For instance, the empire-building argument should be far more relevant to bidding firms with dispersed ownership and entrenched managers than to firms in which the ultimate decision maker owns a sizeable stake (Caprio et al., 2011).

2.3. Diversification

There are several theories that explain why companies do acquisitions and diversify. A com- mon one is the resource-based view, which relates to economics of scope, using excess capac- ity of resources. The assistance of existing (corporate) structures which can be used and are the perk. Further, the market power argument arises from either possibility to gain market share by either using one segment of the company subsidize another and allowing lower pric- es, colluding with other firms over multiple sectors, or squeezing out smaller competitors (Martin, Sayrak 39ff). Diversification, however, increases coordination costs and information asymmetries.

The ability of owners to leverage their own resources and capabilities in shaping a firm´s stra- tegic development hinges on the combination of agency theory with the resource dependency theory, showing how resources affect the impact of boards of directors (Hillman and Dalziel, 2003). Hautz et al. (2013) further extend this approach to the impact of owners on corporate strategy: owners function not only as providers of capital but can contribute a wider set of resources and capabilities that support and facilitate diversification. Differences in the re-

2 The “alignment-of-interest effect” describes the situation that one shareholder holds a great number of shares and therefore the managerial control is low and hence less acquisitions are enabled, and the agency costs are low.

source and capability profiles between different types of owners can thereby affect the rela- tive strategic direction taken by firms. Resources provided by owners may include informa- tional advantages associated with financial institutions (Yuan et al., 2009), expertise in man- aging critical environmental contingencies (Kang and Sorensen, 1999) and access to relational capital (Hautz et al., 2013).

More recent studies on the relationship between ownership and diversification typically take the principal-agent conflict of interest as their point of departure. Prior studies, focusing pri- marily on product diversification, have suggested that managers engage in diversification.

According to the agency theory, the manager wants to diversify due to intrinsic motivation, which can have a variety of causes. One reason can be that his compensation is linked to the company’s success and the manager wants to reduce variability in earnings (Schleifer and Vishny, 1997; Hautz et al., 2013; Aggarwal and Samwick, 2003). Another reason can simply be to attain power and prestige (Aggarwal and Samwick, 2003) or to secure his job due to projects depending on his experience. Lastly, he may want to reduce his personal risk by di- versification (Amihud, Lev 1981).

Creditors could also appreciate the diversification. They have the motivation for more stable cash flows resulting in less delays in payment and liquidity problems. In the situation of high managerial ownership, the shareholder costs of diversification exceed the benefits, so it is less likely that they would support diversification. In contrast, low managerial ownership may facilitate the value-reducing diversification because managers may derive private benefits that exceed their private costs (Denis et al., 1997). Managerial ownership is suggested to support the alignment of incentives between managers and owners whilst concentrated ownership is seen to both incentivize and facilitate the effective monitoring and control of managers (Ami- hud and Lev, 1999).

Campa and Kedia (2002) find that the firm´s decision to refocus can be better explained than the decision to diversify. More specifically, firms with historically high average investments (average capital expenditure (CAPEX) /sales) and low recent investments, are less likely to operate in multiple segments and are more likely to refocus. Also, firms with high historic and recent profitability are more likely to refocus. Firms with high historical average value of as- sets are more likely to operate in multiple segments.

Argumentations on why firms diversify into another sector or country are commonly based on

the transaction cost economics (TCE), which arose in the 70s and 80s and the resource based

view (RBV). The TCE aims to answer the make-or-buy decision (vertical integration), which

means in other words whether the firm diversifies into a new sector or if it sources out (Klein

and Lien, 2009). Based on where resources are useful, the RBV tries to explain into which

sector or country firms diversify. It should, however, be clarified, that the RBV is not limited

to the boundaries of the firm and includes e.g. joint ventures (Klein and Lien, 2009).

2.4. The Nature of Owner

Different shareholders have different incentives to perform acquisitions. According to Hautz et al. (2013), there are three factors that shape the relationship between shareholdings of dif- ferent types of owners and their impact on product and international diversification: motiva- tion, the ability to leverage capabilities to resources and the ability to enforce the owners´ will.

Independent corporations with long-term investments will specialize in monitoring rather than trading (Chen et al., 2007). The presence of these types of institutional shareholders is associ- ated with post-merger performance, which is for the shareholder value in long-term. Families are thought to be focused on the survival of the firm, whilst retaining control, with the ulti- mate goal of passing the firm to descendants (Fiss and Zajac, 2004, p.509).

Financial institutions that normally focus on the adaption of strategies and policies are most likely to optimize firm level economic outcomes (Thomsen and Pedersen, 2000). The ability to monitor and control management to align interests in this context has mainly been stressed by the literature regarding the informational and analytical advantages of financial institutions (Yuan et al., 2009).

The interests of the state as shareholder are of wider political, social and economic outcomes (Schleifer and Vishny, 1997). A government´s incentive may not be to maximize the value of a company, or even to expropriate wealth from minority shareholders, but to provide em- ployment for citizens. When a government seeks to maximize social stability and employment is implemented through a state-owned enterprise (SOE), its form is likely to suffer significant agency costs (Fogel et al., 2008). Therefore, a single controlling shareholder would not be beneficial for the firm if that shareholder is a government owner (Jiang et al., 2017).

Furthermore, the owners proportion of wealth in stake is crucial. An owner who invested a substantial proportion of his wealth will foster firm level diversification to diversify their in- vestments. Compared with governments and financial institutions, families usually invest a much more significant proportion of their wealth in a particular organization and therefore have a higher interest in diversifying acquisitions because they want to diversify their wealth and thereby risk (Miller et al., 2010).

Frydman et al. (1999) argue that, for the setting of a Chinese company, foreign investors may

have managerial know-how advantages that can lead to better firm performance. That is, their

interests may align with minority shareholders and they have specialized knowledge and ex-

perience to help foster it. Also, foreign investors are often associated with better firm-level

corporate governance (Ferreira, Matos, 2008). As these arguments arise from Jiang et al.´s

China based research, the opposite may be found for European companies. For China, foreign

investors, which are non-European and non-US in this study, may bring problems of less

managerial and corporate governance knowledge and experience. Further, Jiang et al. find

that similar shareholders have similar incentives, which are better to align.

3. Hypotheses development

In this section, hypotheses will be developed based on the prior analyzed corporate finance theories with regard to the impact of the shareholder structure on the acquisition behavior.

3.1. Linkage between managerial control and the number of acquisitions

Acquisitions are found to be shareholder value decreasing. Therefore, the shareholders gener- ally want to suppress acquisitions. To find evidence for this argumentation, first the manage- ments´ attitude towards acquisitions will be analyzed to set apart the motivations. In chapter 2.3, multiple possible motivations for the management have been illustrated on why they may favor acquisitions. The separation of ownership and control is a crucial determining factor for how many acquisitions are performed according to Tirole (2006, pp 388). Specifically, com- panies with a strong balance sheet tend to separate ownership and control more than compa- nies with a weak balance sheet. In this context, the strength of a balance sheet mainly depends on relative amount of operating assets to the total assets, the free cash flow to sales and the market interest rate. High liquidity and low market interest rates loosen the tension on the incentive schemes, which then leads to higher managerial control; thus, resulting in more ac- quisitions (Tirole, 2006, pp 388, 401).

In contrast to the prior argumentation, Denis et al. (1997) argue that in the situation of high managerial control, the shareholders are less likely to support diversification because its costs exceed its benefits for themselves. Vice versa, low managerial control may facilitate the val- ue-reducing diversification because managers may derive private benefits that exceed their private costs (Denis et al., 1997). However, this paper will focus on the argumentation of Tirole (2006, pp 388, 401).

H1: Companies with high managerial control perform more diversifying acquisitions than companies with low managerial control.

3.2. Linkage between shareholder size and the number of acquisitions

The management cannot acquire a company without the permission of the supervisory board and hence the large shareholders. Therefore, the shareholders´ motivation will be analyzed.

Firstly, different aspects of the agency theory will be tested in the context of acquisition activ- ity, to set a base upon which more detailed tests can be build up. Whether the proposition by Schleifer and Vishny (1985, 1997), which means in the context of M&A less acquisitions, or the contrary proposition by La Porta et al. can be empirically proven in the context of inor- ganic corporate diversification via acquisitions (linkage between agency costs and acquisition activity) leading to the second hypothesis:

H2: Companies with large shareholders perform less acquisitions than companies without

large shareholders.

Continuing with the testing of the corporate finance theory development in this context, as described in chapter 2.1, the second hypothesis aims to test the later findings that the optimal compromise is to have multiple large shareholders to address both agency issues (Attig et al., 2008; Laeven and Levine, 2008; Bennedsen and Wolfenzon, 2000). More concrete, this paper examines whether large shareholders may collude together against the small shareholders (LaPorta et al., 1999; Bennedsen and Wolfenzon, 2000) and/or if high collaboration and bar- gaining costs between large shareholders may arise (Gomes and Novaes, 2006) and these higher agency costs lead to less acquisitions. Since the findings in the literature on perfor- mance have been diverse, it is also the purpose here to see if performance is as diverse in the context of acquisitions.

H3: Companies with single large shareholders perform more acquisitions than companies with multiple large shareholders.

To complete the hypotheses development on just the acquisition activity, the finding that if the large shareholder has a low level of ownership, the agency costs are high through the “en- trenchment effect” as well as the finding that if the large shareholder has a high level of own- ership, the agency costs are low due to the “alignment-of-interest effect” as the shareholder benefits more from reducing the agency costs by Jensen and Meckling (1976) and Claessens et al. (2002) will be tested as well.

H4: Companies with large controlling shareholders perform more acquisitions than com- panies with large non-controlling shareholders.

Therefore, large shareholders will be further divided into sub-groups according to their share-

holdings to distinguish between large controlling shareholders and large non-controlling

shareholders.

4. Data and Methodology

The following section summarizes the identification methods used to test the hypotheses, in- troduces indicators on how to measure diversification and acquisitions based on observational data, and discusses additional variables that might influence the diversification and acquisi- tion behavior.

4.1. Sample selection and data description

The sample for this investigation is based on all companies included in the “Stoxx Europe TMI Industrials”

index on December 31

, 2016. The index includes a mix of 262 major Eu- ropean industrial companies.

Firstly, common and preferred stocks of the same company are merged in contrast to the sepa- rate view provided by the original list to avoid double counting. Then, companies with insuf- ficient data for the main variables from Orbis

have been excluded. Appendix A provides de- scriptive statistics of the data set after all filters have been applied. This resulted in seven companies being excluded, in a sample including 1,771 firm-year observations representing the 255 unique firms.

Since the continuous company-level variables show some significant dispersion around the mean, a winsorizing technique is used to smooth out the distribution of the variables and to ease the effect of outliers (Bollinger and Chandra, 2005). Thereby, extreme values of both tails are replaced by realization of a predefined percentile of the distribution. In this case, the data has been winsorized by 10%, according to which 5% of the lower and upper tail are re- placed with the realization at the 5% percentile for the lower tail and the 95% percentile for the upper tail respectively. This method is preferred, since the data set is relatively small al- ready and outliers are mainly yearly exceptions, rather than driven by single companies for consecutive years. Thus, the winsorizing is preferred over the mere trimming of data, since it retains more information of the initial data set and accounts for heavily skewed distributions due to overly extreme realizations. After winsorizing, the sample includes 1736 firm-year observations.

The sample includes all acquisitions announced by the sample companies between January 1, 2010 and December 31, 2016. As a sample selection criterion, only acquisitions that are in- cluded in the Zephyr

database are selected. Furthermore, only deals for which the acquirer held less than 50% before deal announcement and obtained majority control of the target firm as result. Further, company data for the control variables is used from the Orbis database as well as the Bloomberg database.

3 https://www.stoxx.com/index-details?symbol=TIDUV

4 Worldwide company information database by the Bureau van Dijk, containing information on over 200 million companies. https://www.bvdinfo.com/en-us/our-products/company-information/international-products/orbis 5Worldwide deal database with integrated company information by the Bureau van Dijk.

https://www.bvdinfo.com/en-gb/our-products/data/specialist/zephyr

Following previous studies (Maury and Pajuste, 2005; Laeven and Levine, 2008; Attig et al., 2008; Attig et al., 2009), a “large shareholder” will be defined as any shareholder who holds 10% or more of the firms' outstanding shares.

Since the historical operating sectors are not available for each company, it is assumed for this research that the major industry classifications did not change over the observed years and always the differing, acquired sectors will be considered. It is very unlikely, that the major sector changes and hence there are no larger implications of this assumption.

4.2. Diversification variables

To have more differentiating models, the dependent variables, describing the company´s di- versification, are being grouped in accordance to the level of detail (Table 1).

Generally, the number of deals (DEALS) will be used to assess the pure acquisition activity and willingness to do an acquisition. Thereby, the further diversification measures can also be used to calculate and interpret the delta to the overall acquisition behavior, separating the in- cremental likelihood or willingness to diversify. This base measure is referred to as “level 0”

in this study.

As next step, the number of acquisitions, with diversifying nature is assessed (DIV). This var- iable can be understood in terms of an and/or condition, meaning acquisitions diversifying in multiple directions as well as acquisitions diversifying in only one direction are inherent of this variable. This next layer is referred to as “level I” diversification in this study.

Table 1

Classification of the dependent variables in levels of detail.

Level Dependent variables

0 DEALS

Le ve l of de ta il

I DEALS_DIV

II

DEALS_DIV_SEC DEALS_DIV_GEO

III

DEALS_DIV_ONLYSEC DEALS_DIV_ONLYGEO DEALS_DIV_SECGEO

IV

DEALS_DIV_GEO_EU

DEALS_DIV_GEO_RoW

To further consider how the companies diversify, geographical (GEO) and sectoral diversifi- cation (SEC) is then separated in level II. Level II only tells if an acquisition has been diversi- fying geographically or sectoral, it does not contain information whether it is a one-directional diversification or two-directional diversification (Figure 1). Distinguishing between only geo- graphical (ONLYGEO) or only sectoral diversification (ONLYSEC) and the combined situa- tion (SECGEO) is done in level III diversification.

Figure 1

Structural model of the diversification (dependent) variable analysis.

The most detailed level (IV) in this study is the distinction between acquisitions which are geographical diversifying, but remain within Europe (EU) and those who are made outside of Europe (RoW). Similarly, the sectoral diversification has been investigated on three different levels of NACE

industry classification hierarchy. This is done for the 4-digit-level (NACE4), 2-digit-level (NACE2) as well as the most aggregated level (noted in letters), summarizing multiple 2-digit sectors (NACE). However, in this study only the 2-digit NACE level will be used since there is not enough variation on NACE4 level. For this research it is not essential if the company is printing books or papers (NACE4), it is however relevant if the company pro- duces the paper or if it prints books (NACE2) The most aggregated form (NACE) on the oth-

6 NACE is the system for statistical classification of economic activities in the European community. The abbre- viation stems from the French term "nomenclature statistique des activités économiques dans la Communauté européenne". The current version form 2008 is revision 2.

er hand combines which are fundamentally different in their products and business models.

Sector code G for example includes whole sale as well as the repair of motor vehicles.

It is important to look at the distribution within the sample in terms of diversification, coun- tries and sectors. The level of sectoral classification is especially relevant, as it is an adjusta- ble factor. The sensitivity will be illustrated in the following paragraph. The level for sectoral diversification has a noticeable influence on the distribution within the sample. While the two-directional diversification (SECGEO) share for the more aggregated NACE level (left) is 21% due to the higher level of detail, for NACE2 the share is more than half of the observa- tions (55%). Accordingly, the observations classified as not being diversifying in any means reduces to a third from 15% to 5%. Appendix L only aims to visualize and make aware of the sensitivity of the variables.

4.3. Ownership variables

As explanatory variables, ownership describing variables are being used. In reference to Jiang et al. (2017), companies will be classified with regard to relative influence (shareholdings in %) as well as control (majority). The first block of independent variables regarding the size of shares hold by each investor. According to the prior mentioned classification of large shareholders holding more than 10% of the shares, this variable equals one if the company in that year has at least one large shareholder (LargeSH). Constitutively, it is distinguished by the quantity; companies with only one large shareholder (SingleLargeSH) and those with more than one large shareholder (MultipleLargeSH). As general control over the company is crucial, it is distinguished if the large shareholder holds more (LargeControllingSH) or less (LargeNonControllingSH) than 50% of the shares, which is still a binary observation.

Large minority shareholders often decide for the group of smaller ones, which leads to a sepa- ration of ownership (formal control) and control (real control by the managers) (Tirole, 2006, 399). Seeing the partly contrary argumentation by Denis et al. (1997) and Tirole (2006), it is crucial for this study to dig deeper and further distinguish between combinations of share- holders as these theories base upon assumptions and in this context, need to be tested on ap- plicability for a subset of shareholder structures.

As shareholdings below 3% are not individually tracked in the databases, three to ten percent (SH3_10) of the shares represents the small shareholders in this sample. The equivalent to

“LargeSH” is “SH10” and “SH50” being about “LargeControllingSH”. The “LargeNonCon- trollingSH” is further divided into “SH10_20” and “SH20_50”. This aims to check whether there is an effect of the at-equity accounting.

Besides the sole size of shareholdings, the nature of the holder is examined is well. The first

group of shareholders are the non-financials, industrial shareholders (OWN_NAT_INDU),

individuals (OWN_NAT_INDI) and the government (OWN_NAT_GOV). The second group

are the financials such as being funds, which are mainly pension funds (OWN_NAT_FUND),

financial institutions (OWN_NAT_FIN) and venture capital as well as private equity firms combined (OWN_NAT_VCPE).

Additionally, an Herfindahl based index of the shareholder diversity with regard to the geog- raphy (OWN_GEO_HER) and their nature such as governmental or financial for example (OWN_NAT_HER). The index is calculated as one minus the sum of the squared values of shareholdings per country and nature, respectively, as a fraction of total shareholdings:

𝐻𝑒𝑟𝑓𝑖𝑛𝑑𝑎ℎ𝑙 = 1 − (∑

𝑝𝑖²) (1)

Accordingly, if 100% of the shares are held by companies from one country, the index equals 0 for “OWN_GEO_HER” and if the shares are held in 10 different countries the index equals

Figure 2

Structural model of the ownership (independent) variable analysis.

0.9. Therefore, often the inverse is used (subtracting the Index form 1 to have an increasing

measure as the diversification increases) (Lang, Stulz 1994). Let p

be the shareholding frac-

tion of the ith country/nature of the company´s total shares and N the number of coun-

tries/natures. Management control is measured according to Tirole (2006, pp 388). As his

argumentation leaves some room for the interpretation on how exactly it should be measure,

two different approaches will be used to calculate a variable in reference to the management

control. The first is based on the operating assets as percentage of the total assets

(OWN_MGN_OA), the other is based on the free cash flow (FCF) as percentage of sales (OWN_MGN_FCF).

4.4. Control variables

To address any concern regarding the omitted variable bias, several company-specific, eco- nomic, and structural control variables are introduced, which might affect the acquisition be- havior. More specifically, to account for economic influences on corporate acquisition deci- sions, two economic indicators are considered. In this regard, the acquisition activity is as- sumed to be lower in times of recession, since companies tend to focus on their core business and tighten their budgets. Therefore, the growth of GDP (GDP_Growth) is included as a con- trol variable to reflect effects of business cycle fluctuations, because companies are expected to adjust their operations. Further exogenous control variable to account for diversification is an indicator tracking the overall M&A activity per year (DEALS_YEAR). If these are found to be significant, the observed correlation between shareholder structure and diversification may not be causal; instead the outcome of actions by profit-maximizing firms reacting to shocks in their environments (Campa and Kedia, 2002).

Further control variables are the natural logarithms of the revenue (SIZE) and the total assets (BS_TOT_ASSET) to account for the size of the companies. Also dummy variables for the countries (CountryX) and the sectors (SectorX), the natural logarithms of the company age (AGE), the leverage of the company (LEV) and the log of the company´s working capital (WCAP), log of the operating assets (OperatingAssets) and the natural logarithms of the free cash flow (FCF) (Denis et al., 1997).

4.5. Model specification and estimation methodology

Panel data is the ideal type of measuring the firm´s acquisition behavior over time. The selec- tive use of fixed effects avoid endogeneity in the error term and allows the use of OLS models estimation. The sample used for this research is an unbalanced longitudinal data panel. Since fixed effects models require longitudinal variation in the data, and some companies in the sample don´t change their major shareholders or industry affiliation over time, whether or not fixed effects can be used will be evaluated for each model separately. According to the out- come of the Hausman test, random or fixed effects will be applied. A statistically significant p-value indicates that there may be a systematic difference in the coefficients, and therefore the fixed-effects ("within") model is preferred rather than include the random-effects ("be- tween") model. Additionally, random effects logit models with the marginal maximum likeli- hood method are employed as robustness checks.

To account for the possibility of heteroskedasticity in the error distribution, robust standard

errors to correct for heteroskedasticity will be used. So, as to test the hypotheses, the identifi-

cation process underlies the following empirical model:

ACQ

= 𝛼 + 𝛽 Σ OWN

+ γ Σ CONTROL

+ δ Σ COMPANY

+ ψ Σ YEAR

+ μ

+ 𝜀

(2) DIV

= 𝛼 + 𝛽 Σ OWN

+ γ Σ CONTROL

+ δ Σ COMPANY

+ ψ Σ YEAR

+ μ

+ 𝜀

(3)

with i = 1, …, n, t = 1, …, T

where measures of corporate acquisition activity (ACQ

) or corporate diversity (DIV

) are used as dependent variable (discussed in 4.2) and the explanatory variables (OWN

) contain a measure of the ownership structure (discussed in 4.3). Additionally, control variables (CONTROL

) are included (discussed in section 4.4), as well as the error term (𝜀

).

4.6. Summary statistics

First, it is important to look at the distribution within the sample in terms of diversification, countries and sectors. The number of acquisitions by type of diversification per year are shown in Table 3, while Figure 3 is the visualization of these. This gives a good overview of this sample´s characteristics regarding the acquisition driven diversification. The overall number of deals follows an upwards sloped S-shape around the mean of 284 acquisitions per year. The low of 259 acquisitions in 2013 is followed by the high two years later of 308 ac- quisitions. Less 2-directional acquisitions (sectoral and geographical diversification) in one year are followed by overall less acquisitions made in the following year. Further, less acqui- sitions come along with more differentiated acquisitions, which are either geographical or sectoral. Absolute and relative increase of acquisitions within Europe over the estimation pe- riod.

Overall, the kind or direction of the acquisitions, relative to the total number per acquisitions, did not differ much over the years.

Figure 3

Line Graph of the number of acquisitions by type of diversification per year (original sample before winsorizing).

Over the years observed in this study, the relative diversification direction of the acquisitions besides less acquisitions outside of Europe and slight tendency towards one-directional diver- sifications, especially only sectoral diversifying. Next aim is to check if there are more differ- ences between sectors and countries with this regard. Due to the unbalanced distribution of the observations over the sectors, in the following, relative values as percentage of the overall acquisitions will be used. It should be noted, that sectors having little share in the sample may deviate from the mean due to company specific attributes instead of sector specific character- istics.

Table 2

Acquisitions by sample firms. This table breaks down the acquisitions on all diversification levels and years (2010-2016). The number of observations is in firm-years.

As for the relative amount of acquisitions, most sectors do not differ much from the average, two do deviate in either direction. Companies in the information and communication sector (J) diversified with every acquisition and 80% of these have been outside of Europe in a sec- tor, which deviates from their core sector. Sector “E” shows contrary characteristics, as only 81% of the acquisitions are diversifying and only 23% are geographically diversifying. Com- panies in this sector are more likely to diversify sectoral via acquisitions. Also, the one- directional geographical diversification is relatively high in comparison to the other sectors, showing that there is a difference from sector to sector whether one- or two-directional acqui- sitions are being preferred.

When looking at the location of acquiring firms, the greatest outlier of the countries is Italy.

The diversification rates are the lowest and more than half of the acquisition are only sectoral diversifying. As little as 13% of the acquisitions have been conducted outside of Europe, which is unlike the average of the manufacturing sector (majority of Italian companies in this sample operate in the manufacturing sector). Contrary, for Spain 40% of the acquisitions have only been geographical diversifying, which is the highest rather for a country in this sample.

The third, characteristic pattern show Switzerland, Denmark and Ireland with most of the ac- quisitions being sectoral diversifying as well as geographical diversifying. Finland shows the most average characteristics around the mean, other countries like Austria have only one at- tribute, which differs from the average more extreme, which is a high tendency towards ac- quisitions within Europe.

Year 2010 2011 2012 2013 2014 2015 2016 Sum

Acquisitions 279 288 266 259 299 308 292 1,991

Diversifying acquisitions 269 267 255 248 281 291 280 1,891

Sectoral diversifying acquisitions 226 216 203 210 243 247 236 1,581

Geographical diversifying acquisitions 191 198 189 176 217 221 205 1,397

Sectoral and geographical diversifying acquisitions 148 147 137 138 179 177 161 1,087

Only sectoral diversifying acquisitions 78 69 66 72 64 70 75 494

Only geographical diversifying acquisitions 43 51 52 38 38 44 44 310

Geographical diversifying acquisitions (EU) 86 89 81 82 94 110 110 652

Geographical diversifying acquisitions (RoW) 105 109 108 94 123 111 95 745

Table 3

Acquisitions by sample firms. This table breaks down the acquisitions made by the sample firms during 2010- 2016 according to their sector. The lower part displays the relative value of diversifying acquisitions to overall acquisitions on diversification level 0-IV for all countries in the sample. The number of observations is in firm- years.

Descriptive statistics for the winsorized company-level data set used throughout the empirical analysis are presented in Table 5 and in Table 6 for the subset including only the observations with at least one deal, while Table 7 and Table 8 comprises the correlation coefficients for respective variables. Conveniently, the average amount of acquisitions a company in the sam- ple made per year is very close to one (0.999), so that incremental impacts of diversification in relation to the acquisition activity can be interpreted more easily. Coming to the description of the binary ownership variables, 32.8% of the companies have one large shareholder, while 39.4% have multiple large shareholder. Further, only 15.7% of the companies are controlled by a shareholder with more than 50% of the shares.

Table 4

Acquisitions by sample firms. This table breaks down the acquisitions made by the sample firms during 2010- 2016 according to their country. The lower part displays the relative value of diversifying acquisitions to overall acquisitions on diversification level 0-IV for all countries in the sample. The number of observations is in firm- years.

Sector C E F G H J K M N O R Sum/Avg.

Acquisitions 931 48 115 215 85 20 26 361 182 0 8 1,991

number of acquisitions

Diversifying acquisitions 894 39 108 205 78 20 25 343 172 0 7 1,891

Sectoral diversifying acquisitions 772 33 102 155 59 16 22 292 124 0 6 1,581

Geographical diversifying acquisitions 748 11 46 137 52 20 12 225 141 0 5 1,397

Sectoral and geographical diversifying acquisitions 626 5 40 87 33 16 9 174 93 0 4 1,087

Only sectoral diversifying acquisitions 146 28 62 68 26 0 13 118 31 0 2 494

Only geographical diversifying acquisitions 122 6 6 50 19 4 3 51 48 0 1 310

Geographical diversifying acquisitions (EU) 345 2 36 91 29 4 3 79 58 0 5 652

Geographical diversifying acquisitions (RoW) 403 9 10 46 23 16 9 146 83 0 0 745

in % of acquisitions

Diversifying acquisitions 96% 81% 94% 95% 92% 100% 96% 95% 95% - 88% 95%

Sectoral diversifying acquisitions 83% 69% 89% 72% 69% 80% 85% 81% 68% - 75% 79%

Geographical diversifying acquisitions 80% 23% 40% 64% 61% 100% 46% 62% 77% - 63% 70%

Sectoral and geographical diversifying acquisitions 67% 10% 35% 40% 39% 80% 35% 48% 51% - 50% 55%

Only sectoral diversifying acquisitions 16% 58% 54% 32% 31% 0% 50% 33% 17% - 25% 25%

Only geographical diversifying acquisitions 13% 13% 5% 23% 22% 20% 12% 14% 26% - 13% 16%

Geographical diversifying acquisitions (EU) 37% 4% 31% 42% 34% 20% 12% 22% 32% - 63% 33%

Geographical diversifying acquisitions (RoW) 43% 19% 9% 21% 27% 80% 35% 40% 46% - 0% 37%

Country AT BE CH DE DK ES FI FR GB IE IT NL NO PT SE Sum/Avg.

Acquisitions 58 24 190 119 36 45 175 212 539 31 30 43 49 0 440 1,991

number of acquisitions

Diversifying acquisitions 55 22 187 108 36 42 164 209 506 31 24 41 43 0 423 1,891

Sectoral diversifying acquisitions 45 18 175 103 29 24 135 177 400 30 20 35 40 0 350 1,581

Geographical diversifying acquisitions 44 13 175 82 36 36 123 168 319 30 8 26 22 0 315 1,397

Sectoral and geographical diversifying acquisitions 34 9 163 77 29 18 94 136 213 29 4 20 19 0 242 1,087

Only sectoral diversifying acquisitions 11 9 12 26 0 6 41 41 187 1 16 15 21 0 108 494

Only geographical diversifying acquisitions 10 4 12 5 7 18 29 32 106 1 4 6 3 0 73 310

Geographical diversifying acquisitions (EU) 32 7 72 43 15 14 66 50 100 17 4 14 17 0 201 652

Geographical diversifying acquisitions (RoW) 12 6 103 39 21 22 57 118 219 13 4 12 5 0 114 745

in % of acquisitions

Diversifying acquisitions 95% 92% 98% 91% 100% 93% 94% 99% 94% 100% 80% 95% 88% - 96% 95%

Sectoral diversifying acquisitions 78% 75% 92% 87% 81% 53% 77% 83% 74% 97% 67% 81% 82% - 80% 79%

Geographical diversifying acquisitions 76% 54% 92% 69% 100% 80% 70% 79% 59% 97% 27% 60% 45% - 72% 70%

Sectoral and geographical diversifying acquisitions 59% 38% 86% 65% 81% 40% 54% 64% 40% 94% 13% 47% 39% - 55% 55%

Only sectoral diversifying acquisitions 19% 38% 6% 22% 0% 13% 23% 19% 35% 3% 53% 35% 43% - 25% 25%

Only geographical diversifying acquisitions 17% 17% 6% 4% 19% 40% 17% 15% 20% 3% 13% 14% 6% - 17% 16%

Geographical diversifying acquisitions (EU) 55% 29% 38% 36% 42% 31% 38% 24% 19% 55% 13% 33% 35% - 46% 33%

Geographical diversifying acquisitions (RoW) 21% 25% 54% 33% 58% 49% 33% 56% 41% 42% 13% 28% 10% - 26% 37%

Industrial companies as shareholder make up 13.8%, individuals account for 3.3% and the

government holds only 1.4% of the shares in the sample. More dominant are funds (14.2%)

and financial institutions (16.4%), whereas venture capital and private equity firms only hold

3.2%. The coverage of the total shares, which is analyzed by nature in the study, therefore

equals 52.3%. The shareholder dispersity is measured with the Herfindahl indices, which

mean that on average, the shareholders (holdings >3% are considered) are from 2.66 different

countries and have 2.15 different natures. The average company in the sample has been listed

41 years ago. For the sample group, the average year on year GDP growth is 1.32%, which is

the weighted average over the sample.

Table 5

Descriptive statistics of sample companies. This table presents descriptive statistics of the 248 sample companies.

The diversification variables are described in chapter 4.2, ownership variables in chapter 4.3 and control varia- bles in 4.3. All variables are winsorized at 0.05 and 0.95. The number of observations is in firm-years.

FULL SAMPLE Observations Mean Standard

Deviation Minimum Maximum Diversification / Dependent variables

DEALS 1,736 0.999 1.421 0.000 5.000

DEALS_DIV 1,736 0.956 1.395 0.000 5.000

DEALS_DIV_SEC 1,736 0.777 1.157 0.000 4.000

DEALS_DIV_GEO 1,736 0.698 1.137 0.000 4.000

DEALS_DIV_GEO_EU 1,736 0.310 0.615 0.000 2.000

DEALS_DIV_GEO_RoW 1,736 0.321 0.633 0.000 2.000

DEALS_DIV_SECGEO 1,736 0.525 0.891 0.000 3.000

DEALS_DIV_ONLYGEO 1,736 0.112 0.315 0.000 1.000

DEALS_DIV_ONLYSEC 1,736 0.173 0.378 0.000 1.000

Ownership / Independent variables

LargeSH binary 1,651 0.721 0.448 0.000 1.000

SingleLargeSH binary 1,651 0.328 0.470 0.000 1.000

MultipleLargeSH binary 1,651 0.394 0.489 0.000 1.000

LargeControllingSH binary 1,651 0.157 0.364 0.000 1.000

LargeNonControllingSH binary 1,651 0.572 0.495 0.000 1.000

SH3_10 1,651 4.801 4.130 0.000 13.000

SH10_20 1,651 0.686 0.943 0.000 3.000

SH20_50 1,651 0.452 0.683 0.000 2.000

SH50 1,651 0.157 0.364 0.000 1.000

SH3_20 1,651 6.239 4.087 1.000 15.000

SH10 1,651 1.345 1.136 0.000 4.000

SH20 1,651 0.603 0.713 0.000 2.000

OWN_NAT_INDU 1,651 0.138 0.197 0.000 0.657

OWN_NAT_INDI 1,651 0.033 0.088 0.000 0.345

OWN_NAT_GOV 1,651 0.014 0.030 0.000 0.106

OWN_NAT_FUND 1,651 0.142 0.151 0.000 0.512

OWN_NAT_FIN 1,651 0.164 0.150 0.000 0.519

OWN_NAT_VCPE 1,651 0.032 0.061 0.000 0.223

OWN_MGN_OA 1,689 0.351 0.242 0.000 0.774

OWN_MGN_FCF 1,671 0.102 0.091 0.000 0.322

Control variables

AGE 1,771 32.049 34.089 0.000 114.000

SIZESALES 1,688 8,056.081 12,596.330 305.800 48,582.000

BS_TOT_ASSET 1,690 9,042.256 13,202.710 425.000 48,401.300

LEVERAGE 1,771 0.953 0.802 0.000 3.082

WCAP 1,771 579.744 869.937 -18.400 3,314.592

OperatingAssets 1,771 2,312.390 3,616.391 0.000 14,121.000

FCF 1,771 384.781 514.844 0.000 1,992.000

GDP_Growth 1,771 1.565 1.215 -1.057 3.842

DEALS_YEAR 1,771 284.429 16.278 259.000 308.000